Pattern Classification Using Secondary Components Perceptron and Economic Applications

In this paper we will classify patterns using a modified Perceptron algorithm (Dumitrache et al., 1999). The generalization uses the eigenvalues and the eigenvectors of the sample covariance matrix, as we did for classifying patterns using PCR (Ciuiu 2007b).We shall also define measurements for the cohesion of the obtained classes and of the separation between them. The first economic application considered in the paper is a consumer behavior model (Jula 2003), and the second is the same financial application for classifying banks (Ciuiu, 2007a, Ciuiu, 2007b), where we have used regression for classification.